AI Companies Tread Legal Thin Ice: Copyright and 'Fair Use' in Focus

Introduction

In the rapidly evolving landscape of artificial intelligence, the intersection of AI and intellectual property rights poses a significant challenge. The rise of AI has sparked innovation across various sectors, yet it has also ushered in complex legal dilemmas, particularly around the use of publicly available content. A critical misunderstanding persists in the realm of AI training—that content visible online can be freely used. This assumption has led many AI companies to utilize a vast array of publicly accessible data, sometimes without considering copyright implications. As highlighted in the article, this misconception can lead to legal risks for companies in the AI sector [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

A key aspect of this issue is the "fair use" doctrine, which is often invoked to justify the inclusion of copyrighted materials in AI databases. However, the doctrine's application is nuanced and context-dependent, involving a four-factor test to evaluate whether a particular use qualifies as "fair." This legal concept has become a focal point of debate, as it was prominently featured in cases like *Thomson Reuters v. Ross Intelligence*. In this case, the court's ruling against Ross Intelligence's fair use defense underscored the limitations and complexities of relying on this doctrine in AI contexts, as detailed in the provided article [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

For AI companies, the integration of copyrighted material into training models without explicit permission from copyright holders not only risks infringement but also poses strategic and financial challenges. Legal battles could result in extensive costs, damages, and the possibility of having to overhaul training data sets. The article stresses the need for AI entities to engage with legal experts, conduct audits, and potentially reshape their strategies to include licensing agreements that respect intellectual property rights [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Moreover, content creators and copyright holders find themselves in a precarious position as AI technologies continue to evolve. The *Thomson Reuters v. Ross Intelligence* decision illustrates the necessity for creators to actively protect their intellectual property and explore licensing opportunities. As the article points out, ensuring creative works are adequately safeguarded against unauthorized use in AI systems reflects a growing concern within the creative industries [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

The ongoing development in AI technology necessitates a balanced approach, where both innovation and legal integrity are maintained. By fostering collaborations between tech innovators, legal experts, and content creators, the industry can work towards a sustainable model that protects intellectual property without stifling technological progress. This is a pivotal moment for stakeholders to redefine legal frameworks to adequately address both current and future challenges in AI development, as highlighted in the outlined article [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Legal Challenges of AI and Copyrighted Content

The intersection of artificial intelligence (AI) and copyright law presents a complex landscape where technological advancement challenges traditional legal frameworks. AI companies often utilize copyrighted content to train machine learning models, raising significant legal concerns regarding intellectual property. Publicly available content, especially that which is copyrighted, is not inherently free to use for AI training purposes. This legal challenge underscores the importance of understanding that publicly visible does not equate to public domain. Recent cases highlight that unauthorized use of copyrighted material can lead to claims of infringement, even if only a subset of that content is utilized in training datasets. Courts are increasingly scrutinizing such practices and ruling on whether they fit within the doctrine of fair use, which remains a contentious legal battleground. For AI companies, this means more than ever, the focus must shift towards comprehensive copyright compliance strategies and possibly exploring viable licensing models to mitigate risks.

The legal doctrine of 'fair use' is central to ongoing debates about AI and copyright. Traditionally, fair use permits limited use of copyrighted works without a license, employing a four-factor test that considers the purpose, nature, amount, and effect of use. However, applying this framework to AI poses unique challenges. AI learning often involves ingesting vast quantities of data—some of which may be copyrighted—and creating outputs based on this data. This has led to contentious discussions on whether such uses are transformative enough to qualify under fair use. Cases like *Thomson Reuters v. Ross Intelligence* illustrate the complexity. In this case, the court found that using a competitor's legal summaries in AI training did not satisfy fair use criteria, especially when it posed a direct market competition. Such rulings spotlight the need for AI companies to navigate these waters carefully, considering both legal guidance and ethical implications.

Copyright issues pertaining to AI involve intricate considerations of intellectual property rights and competitive practices. The legal landscape is dynamic, driven by cases involving major AI technology players, sparking debates on whether AI enhancement strictly involves learning or if it inadvertently mirrors and monetizes existing copyrighted works. While AI systems are lauded for their innovative capacity, the reliance on copyrighted content without explicit permission challenges the boundaries of copyright law and its application to AI. A sweeping conclusion cannot cover the intricate legalities involved, but each case contributes to a developing body of law that will guide future AI applications. As a result, AI entities must adopt proactive measures, such as thorough auditing of data sources and exploring licensing agreements, to ensure compliance and to prevent costly litigation. Both content creators and AI developers must engage in ongoing dialogue and negotiation to balance innovation with respect to intellectual property rights.

As AI technology continues to advance, the precedent set by copyright-related rulings will have significant implications for both AI developers and content creators. Legal actions, such as the *Thomson Reuters v. Ross Intelligence* lawsuit, where the court dismissed claims of fair use, indicate that the unchecked use of copyrighted material is legally fraught. This adds pressure on AI firms to ensure their training data complies with copyright laws. Concurrently, content creators are pushed to safeguard their works against unauthorized usage. They must remain vigilant and proactive in monitoring how their content is utilized, while also considering entering into licensing agreements that could offer revenue streams from AI companies seeking to legitimize their training datasets. Both parties stand to benefit from transparent communication channels and the establishment of industry-wide standards for copyright compliance.

Understanding 'Fair Use' and Its Complexity in AI

"Fair use" is a legal doctrine that balances the rights of copyright owners with the public interest in accessing and utilizing creative works, and its application to artificial intelligence (AI) is a growing area of legal complexity. In the realm of AI, particularly concerning the training of machine learning models, questions arise about whether the use of copyrighted content without explicit permission can be considered "fair use." As AI systems require vast amounts of data for training, the temptation to use publicly available content is significant. However, using such data without proper licensing may not always align with fair use principles, especially when the content is protected by copyright laws.[1]

The complexity of applying fair use to AI is underscored by the four-factor test used by courts to evaluate such claims. This test assesses the purpose and character of the use, the nature of the copyrighted work, the amount of the work used, and the effect on the work's market value. These factors gain particular significance in AI-related cases, where the tools are often used commercially and compete with original works in the marketplace. The judgment in *Thomson Reuters v. Ross Intelligence* highlighted these complexities by ruling against the fair use defense for AI training, emphasizing that publicly available legal summaries were not free of copyright constraints merely because they were accessible online.[1]

The AI industry faces growing challenges in adapting to these legal standards. As noted in discussions about the *Thomson Reuters* case, AI firms must take proactive measures to comply with copyright laws, which may include seeking licenses for data use. This brings into focus the broader debate on the reach of copyright law and how it grapples with technological advancements. Industry observers suggest that AI companies need to be circumspect about the sources of their training data, conducting audits, and adjusting their practices to mitigate litigation risks.[1]

The implications of these legal battles extend beyond single court decisions. They influence the strategies of AI companies globally, urging them to prioritize transparency and legal compliance. The legal landscape around fair use is still evolving, and AI developers must stay informed about changes as more cases are adjudicated. The importance of such awareness is underscored as AI technologies continue to expand into new domains, carrying with them both the promise of innovation and the responsibility of ethical practice. Proper understanding and application of fair use principles are pivotal in navigating these uncharted waters and ensuring the sustainability of AI progress.[1]

Ultimately, the conversation around fair use in AI isn't just a legal issue—it's a philosophical and ethical inquiry into how society values creativity and the fruits of intellectual labor. As AI continues to push boundaries, striking a balance between innovative freedom and respect for existing works remains central to debates within the industry. This is why cases like *Thomson Reuters v. Ross Intelligence* are significant; they reflect ongoing tensions and the need for adaptive legal frameworks capable of responding to the unique challenges posed by AI.[1]

Key Court Cases: Focus on *Thomson Reuters v. Ross Intelligence*

The case of *Thomson Reuters v. Ross Intelligence* is a crucial moment in the intersection of artificial intelligence and copyright law. In this landmark decision, the court refused to recognize Ross Intelligence's claim of 'fair use' for employing copyrighted legal summaries in training its AI systems. This ruling marks a significant precedent, emphasizing that openly available content, particularly those with commercial significance and created by humans, cannot be freely utilized for AI training purposes without potential legal repercussions. This case underscored the critical evaluation of fair use through a four-factor test and suggested that the economic impact and nature of the copyrighted work weighed heavily against Ross [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Moreover, the court's stance in this case reflects an increasing urgency to address how AI companies access and utilize data. By rejecting Ross Intelligence's defense, the court highlighted the potential risks associated with the unlicensed use of copyrighted material, urging AI developers to reconsider their data acquisition strategies. The court's decision emphasizes that even data not included in the final AI outputs can still constitute copyright infringement if used in the training process, thus reshaping how companies approach AI model training and the legal obligations involved [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

The *Thomson Reuters v. Ross Intelligence* case illuminates the broader implications for AI developers who must navigate complex copyright landscapes. As AI technology evolves, so does the scrutiny on how training data — especially that which is copyrighted — is utilized. With this ruling, a clear signal is sent to AI companies, encouraging them to pursue licensing agreements and enhance transparency in their data sourcing processes. The judgment further denotes the necessity for a balanced approach where innovation does not overshadow creators' intellectual property rights [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Consequences of Copyright Infringement for AI Companies

With the rapid evolution of artificial intelligence, AI companies find themselves navigating complex legal landscapes, particularly concerning the use of copyrighted material for model training. As emphasized by the article, the notion that publicly visible content is free for mining by AI companies is proving misleading and legally perilous. Copyright infringement claims are not only a potential financial burden but also tarnish reputations, resulting in legal injunctions that can severely hinder operational workflows [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

In navigating these legal waters, the "fair use" doctrine becomes a contested shield for AI companies attempting to justify their use of copyrighted content. However, as highlighted in cases like *Thomson Reuters v. Ross Intelligence*, courts are reluctant to entertain fair use defenses when the intent behind using such content infringes commercial interests [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking). This reluctance underscores the necessity for AI firms to reevaluate their data acquisition strategies critically.

AI companies may face severe consequences if they forge ahead without adequate copyright compliance. These consequences could range from costly legal battles to the arduous process of scrapping and rebuilding AI models with alternative data sources, as the article points out. Such legal disputes not only drain resources but may also divert focus from innovation to damage control [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

The economic repercussions extend to the potential restructuring of the AI industry itself, where licensing models may become the norm. This shift would require companies to invest in legal expertise and licensing agreements, creating a more sustainable, albeit costlier, path to development. As more AI companies face litigation, the pressure to secure rights to training data legally will likely intensify [1](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Strategies for Navigating Legal Complexities

Navigating legal complexities in the age of AI requires a nuanced understanding of copyright laws, especially as they pertain to publicly available content. The assumption that content visible to the public is free to use is a common misconception that can lead to significant legal challenges. For instance, the use of such content to train AI models often leads to debates over copyright infringement and fair use claims. The legal framework in place, particularly the fair use doctrine, provides some latitude, but its applicability is heavily dependent on specific factors such as the purpose of use and market impact. As illustrated by the Thomson Reuters v. Ross Intelligence case, courts are skeptical of arguments that AI can freely use copyrighted content without due consideration or compensation.

To effectively navigate these complexities, AI companies should consider a proactive legal strategy. This involves a comprehensive audit of their training data to assess potential legal risks. It's crucial for these companies to explore licensing models as a means of avoiding copyright infringement pitfalls. By securing proper licenses, firms can both protect their operations and support the rights of content creators. Furthermore, consulting with legal experts who are versed in intellectual property law can provide tailored guidance. These actions not only mitigate risks but also promote an ethical approach to AI development.

Content creators, on the other hand, must remain vigilant to protect their intellectual property. Monitoring the use of their works in AI training and seeking legal recourse when necessary are essential steps. Encouragingly, the rise in awareness and legal precedents provides content creators with better tools to safeguard their creations. Licensing frameworks offer a viable solution to monetize their content, ensuring they are compensated fairly for its use while allowing AI companies to continue innovating within legal bounds. This synergy between creators and developers promotes a balanced ecosystem where technological advancement and intellectual property rights coexist successfully.

Current Events Impacting AI and Copyright

The rapid advancement of artificial intelligence (AI) has brought significant attention to the intersection of AI and copyright law. One of the major current events affecting this landscape is the ongoing controversy surrounding the use of copyrighted content for training AI models. As highlighted by sources like , the premise that 'publicly visible' content is free for use is a common misconception. This misunderstanding often leads to legal challenges, particularly when AI companies utilize copyrighted material without permission under the guise of 'fair use,' a doctrine that allows limited use of copyrighted material without the owner's consent. The ongoing debate over what constitutes 'fair use' in AI training continues to be a contentious legal issue.

A pivotal case that exemplifies these issues is the *Thomson Reuters v. Ross Intelligence* lawsuit. The court's rejection of Ross Intelligence's fair use defense has set a precedent that publicly available content, especially when commercially valuable and crafted by humans, is not inherently free for AI training. This ruling emphasized the importance of copyright compliance and the potential risks AI companies face, including injunctions and retraining costs, should they infringe upon copyright laws. Such cases have incentivized AI companies to re-evaluate their data usage policies and explore licensing agreements as a means to mitigate potential legal repercussions.

Moreover, the proliferation of lawsuits against AI companies by major news organizations and authors underscores the growing regulatory scrutiny AI technology faces. These legal battles are not just confined to the U.S.; they have international dimensions that reflect the global nature of AI development and data usage. A case from India, , highlights these global challenges, indicating a potential need for coordinated international efforts in shaping copyright laws relevant to AI. Such international cases point to future requirements for a cohesive global approach in addressing AI's legal implications.

On the business front, the economic repercussions for AI companies could be profound. As noted by various experts, compliance with copyright laws could lead to increased operational costs due to the necessity of licensing agreements and possible restructuring efforts to create legally compliant datasets. This scenario might elevate the financial burdens on smaller AI startups trying to compete with established giants who can better absorb these costs. However, it might also open new revenue streams for content creators as licensing becomes a more dominant aspect of AI development. The socio-economic landscape of the AI industry is thus poised for significant transformation in the wake of these legal trails.

Public perception also plays a critical role in shaping how AI copyright issues are addressed. There's a growing call for transparency in the data practices of AI companies. As public awareness increases, so does the pressure on these companies to adopt transparent and ethical data usage policies. The discourse also signals a potential shift in public sentiment, where AI is both critically eyed for its potential legal oversteps and lauded for its innovative capabilities. This complex dynamic between technological progress and regulatory adherence will continue to influence how companies navigate the copyright constraints associated with AI development.

Expert Opinions on Legal Challenges

Experts in the legal field are increasingly voicing their opinions on the challenges associated with using publicly available content for AI training. A common sentiment among these experts is skepticism about the reliance on 'fair use' as a defense by AI companies. The decision in the landmark Thomson Reuters v. Ross Intelligence case, where the fair use defense was rejected, marks a critical point in this ongoing debate. According to experts, this case illustrates that even minimal use of copyrighted material, if intended for commercial gains, may not be justified as fair use. The court meticulously evaluated the impact of such usage on the market potential of the original copyrighted works, further challenging the assumptions AI companies might have about freely available content. More details can be explored in this article on copyright challenges in the AI age.

Another significant perspective comes from the rise of legal scrutiny surrounding AI-related copyright challenges. Experts are advising AI firms to take proactive steps toward mitigating risks associated with their use of copyrighted content. Recommendations include conducting thorough audits of their training datasets, exploring licensing models, and consulting with legal experts to navigate these complex legal waters. Moreover, the notion that the use of copyrighted material in AI training, even when it does not appear in the final output, might still constitute infringement is gaining traction. The emphasis is on a comprehensive understanding of copyright laws and the development of robust compliance mechanisms. For more insights into compliance advice for AI companies, readers might find this source useful: details on AI and copyright.

Public Reactions to AI's Use of Public Data

Public reactions to the use of AI in processing public data reveal the intricate challenges between innovation and intellectual property rights. Many users feel that although the data might be publicly visible, it doesn't translate to being free for unrestricted use. Understanding that there exists a difference between 'publicly available' and 'free to use' is blurry for the general public, particularly when data is involved in complex processes such as AI training. Organizations capitalizing on such content have to tread carefully, for the legal frameworks are still developing and often rest on varied interpretations of 'fair use' as highlighted by cases like *Thomson Reuters v. Ross Intelligence* .

The misconception that anything online is free for any kind of exploitation is increasingly being challenged. Many in the public domain were shocked to realize that AI training methods potentially infringe on individual copyrights, thereby forcing a paradigm shift in how AI developers approach data acquisition. The debate surrounding AI and 'fair use' has sparked meaningful discourse, urging AI companies to reassess how they view copyrighted materials and strategically align their projects with ongoing legal standards. As shown in recent discussions, transparency and honest data sourcing have become pressing concerns among users who fear their own contributions to public spaces being leveraged without consent.

Future Implications of AI Copyright Issues

As artificial intelligence (AI) technologies advance, the intersection between AI and copyright law is becoming increasingly complex, particularly with regard to how AI companies utilize publicly accessible content in training their models. A mistaken belief persists that 'publicly visible' content is free for use, which has significant legal implications when such content is copyrighted. This is reflected in the high-profile *Thomson Reuters v. Ross Intelligence* case, where the court dismissed the 'fair use' defense offered by Ross Intelligence for using copyrighted legal summaries. This legal precedent emphasizes that AI companies must tread carefully when their models access data that appears publicly available but is protected by copyright laws. [Read more about the case and its implications here](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

The debate around 'fair use' in the context of AI is becoming a focal point for legal discussions. The standard four-factor test that determines fair use—considering the purpose and character of the use, the nature of the copyrighted work, the amount and substantiality taken, and the effect of the use on the market—struggles to adapt to AI training methodologies. This complex legal landscape necessitates AI companies to consider their operations critically and adopt proactive copyright compliance strategies. Failure to do so could lead to costly litigation and disrupt the development and deployment of AI models. [Learn more about 'fair use' in the age of AI](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Moreover, the economic implications of rising copyright issues in AI are formidable. AI companies might need to negotiate licensing agreements extensively, which could increase operational costs considerably. This may also slow down the pace of innovation as smaller startups might struggle to afford such licenses, thereby reducing competitiveness in the market. Conversely, this shift could provide new revenue streams to content creators who license their work for AI training purposes. Essentially, a new economy centered around content licensing may emerge, reshaping the dynamics of the AI training landscape. [Discover more about the economic shifts in AI copyright issues](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Economic, Social, and Political Implications

The economic implications of legal disputes over AI training data are multifaceted, potentially reshaping the industry's financial landscape. As highlighted in the *Thomson Reuters v. Ross Intelligence* case, AI companies may need to embrace licensing agreements to avoid copyright infringements. This shift not only increases the cost of developing AI technologies but also benefits content creators by creating new revenue streams. However, these costs may stifle innovation [source].

Uncertainty and the Path Forward for AI Legal Landscape

The current legal environment concerning AI and copyright presents numerous uncertainties. Many companies find themselves caught between the potential for innovation and the constraints of existing intellectual property laws. This uncertainty is especially pronounced in the use of "publicly visible" content to train AI models, where the legal distinction between accessible and usable is often blurred [News Article](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking). The rapid progression of AI technologies surpasses the pace at which legislation can adapt, leaving a gap that legal professionals and stakeholders must navigate. Courts, as seen in landmark cases like *Thomson Reuters v. Ross Intelligence*, are setting precedents that challenge the assumption that visible equates to free to use.

Moving forward, stakeholders in the AI industry are advised to adopt a proactive stance regarding copyright compliance. AI companies must carefully audit their data sources and engage in due diligence to mitigate litigation risks. This includes exploring various licensing models that not only respect the intellectual property of content creators but also provide the necessary resources for AI development to flourish. A notable suggestion is for AI developers to forge partnerships with content creators, ensuring that usage rights are respected and potentially monetized [News Article](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Despite the challenges, there is a path forward that includes collaboration between AI companies, legal experts, and policymakers. Establishing clear guidelines and developing a regulatory framework that keeps pace with technological advancements is imperative. The development of industry standards for data usage and copyright could alleviate mutual concerns and promote transparent practices within the industry [News Article](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking). This should be done while ensuring these standards encourage innovation without compromising the rights of content owners.

The path to a legally secure AI future involves addressing these uncertainties through comprehensive dialogue and strategic policy-making. This could include forming international coalitions that address the cross-border nature of AI development and data regulation. Implementing a consistent global framework could aid in reducing disparities between jurisdictions and provide a predictable legal landscape in which AI innovation can thrive without overstepping copyright laws. As AI technology continues to evolve, so too must the legal frameworks that support its responsible growth [News Article](https://www.mondaq.com/unitedstates/new-technology/1611418/copyright-in-the-age-of-ai-why-publicly-visible-content-isnt-free-for-the-taking).

Conclusion

In conclusion, the rapidly evolving landscape of artificial intelligence presents an array of challenges, particularly concerning the use of publicly visible content for training AI models. Though easily accessible, the assumption that such data is free to use has led to significant legal disputes, underscoring that accessibility does not equate to permission. The pressing need for AI companies to reconsider their strategies is highlighted by lawsuits emphasizing copyright compliance and the exploration of licensing models as prudent steps forward.

Learning from pivotal cases like Thomson Reuters v. Ross Intelligence, it becomes evident that the legal environment will play a decisive role in shaping the future of AI. The rejection of the fair use defense illustrates the emphasis on protecting the commercial value of copyrighted content, even in the face of technological innovation. Consequently, AI developers must navigate this terrain with care, seeking licenses and protecting against infringement claims, which could otherwise result in costly legal entanglements.

Moving forward, the integration of ethically sourced data and proactive legal compliance will be key to sustainable AI development. With numerous stakeholders expressing a vested interest, from content creators worried about intellectual property rights to tech entities aiming for innovation, a balanced approach that respects copyright laws while fostering technological progress is critical. This balance will not only protect creative outputs but also guide AI technologies toward a more compliant future.

Ultimately, the ongoing dialogue between AI companies, legal experts, and content creators presents an opportunity to establish a cooperative framework. Such collaboration can help develop international standards and regulations that ensure fairness and innovation. As the debate on fair use continues, it will be crucial to identify solutions that uphold the integrity of intellectual property while allowing AI technologies to advance responsibly.